Ant Group's Bailing officially open-sourced its next-generation native hybrid reasoning model Ling-3.0-flash yesterday, focusing on extreme intelligence density, a full-chain Agent closed loop, and low-cost large-scale deployment.
Huawei officially announced today that Ascend has simultaneously completed 0-day adaptation for the Bailing Ling-3.0-flash model. During the adaptation process, CANN PyPTO, an operator programming framework, was introduced for the first time, which can shorten the delivery cycle for complex fused operators and efficiently complete operator integration and inference framework performance tuning, while also making the complete deployment project available.
This adaptation is fully compatible with the Ascend A2 and A3 series products and comes with an official standardized deployment solution. Developers can quickly launch inference services through the official image (hardware requirements: Ascend A2 series products: 8 cards; Ascend A3 series products: 4 cards).
The CANN PyPTO programming framework is built on PTO ISA (virtual instruction set) and adopts the Tile programming paradigm. Developers can precisely define computing logic through Tensor-level APIs, while the framework automatically completes the entire process of tiling and scheduling, data movement, and code generation.
In addition, the CANNBot PyPTO Agent has built a multi-agent framework that automatically advances the entire operator development process through 7 stages, with each stage carried out collaboratively by specialized sub-agents. All scheduling is handled by the unified orchestrator pypto-op-orchestrator. The sub-agents exchange information through a memory mechanism and machine-readable states, ensuring seamless context transfer.
The entire process begins with natural-language operator requirements and proceeds through automated generation, verification, and tuning until it produces an on-board-ready operator that meets accuracy requirements and has been performance-optimized. No manual intervention is required for handoffs between stages. This workflow not only delivers a comprehensive upgrade to the operator programming experience but also significantly shortens the operator delivery cycle. The logic of each stage is as follows:
